Download Scientific Uncertainty and the Political Process

American Academy of Political and Social Science
Scientific Uncertainty and the Political Process
Author(s): Dale Jamieson
Source: Annals of the American Academy of Political and Social Science, Vol. 545, Challenges
in Risk Assessment and Risk Management (May, 1996), pp. 35-43
Published by: Sage Publications, Inc. in association with the American Academy of Political
and Social Science
Stable URL: http://www.jstor.org/stable/1047890
Accessed: 05/06/2010 19:36
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=sage.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
Sage Publications, Inc. and American Academy of Political and Social Science are collaborating with JSTOR
to digitize, preserve and extend access to Annals of the American Academy of Political and Social Science.
http://www.jstor.org
ANNALS, AAPSS, 545, May 1996
Scientific Uncertainty and the Political Process
By DALEJAMIESON
ABSTRACT
In this article, a notion of scientific uncertainty is
sketched that is in many ways different from the prevailing view.
Scientific uncertainty is not simply an objective value that can be
reducedby science alone. Rather,scientific uncertainty is constructed
both by science and by society in order to serve certain purposes.
Recognizingthe social role of scientific uncertainty will help us to see
how many of our problems about risk are deeply cultural and cannot
be overcomesimply by the application of more and better science.
Dale Jamieson is professor of philosophy at the University of Colorado, Boulder,
and adjunct scientist in the Environmental and Societal Impacts Group at the National
Center for Atmospheric Research. He received his Ph.D. from the University of North
Carolina at Chapel Hill and previously taught at North Carolina State University and
the State University ofNew York.He has also held visiting positions at Cornell, Oxford,
and Monash University in Australia. He is the editor of five books and has published
many articles on ethics, science policy, and environmental philosophy.
35
36
THE ANNALSOFTHE AMERICANACADEMY
of the most controversial
SOME
public policy decisions in Ameri-
can society involve risks that are primarily understood through scientific
processes and institutions. The evidence for climate change, for example, comes mainly from experiments
run on highly complex climate models rather than from our everyday
experience. Other issues with important scientific dimensions include
ozone depletion, biodiversity loss,
acid rain, and exposure to radon and
various toxic chemicals. Without science and scientists, there would be
little public concern about a wide
range of important issues.
Although science has been very effective in bringing these issues into
the public arena, it has been quite
ineffective at providing solutions.
There are a number of views about
why this is the case. Over lunch and
at professional meetings, scientists
often complain about the lack of understanding or downright perversity
on the part of political leaders who
ignore scientific information. On the
other hand, many policy analysts
fault scientists for talking to each
other rather than producing"policyrelevant" science.1 My own view,
which cannotbe fully developedhere,
is that the very characteristics of science that enable it to have its unique
cultural authority as a knowledge
producer disable it from bringing
public decisions to closure.2
1. See, for example, E. S. Rubin, L. B.
Lave, and M. G. Morgan, "Keeping Climate
Research Relevant," Issues in Science and
Technology, 8(2):47-55 (1991-92).
2. I have developed this view more fully in
a number of papers. See, for example, "Ethics,
Public Policy and Global Warming," Science,
Technology and Human Values, 17(2):139-53
(1992).
The conventional wisdom about
why science is often so ineffective in
providing solutions to problems with
important scientific dimensions focuses on the role of uncertainty. In
this view, problems such as climate
change are characterizedby high levels of scientific uncertainty about the
likelihood and effects of key events,
and so partisans of various policies
can use-or misuse-scientific information and authority for their own
purposes. For example, although the
weight of scientific evidence suggests
that large-scale emissions of greenhouse gases are likely to change climate, there are so many uncertainties about the roles of clouds, carbon
sinks, and various possible feedbacks
that both greenhouse "hawks" and
"doves"can reasonably enlist science
as an ally while accusing their opponents of misusing science.3The only
way out of this situation, some argue,
is for uncertainties to be reduced to
the point at which science can determine a rational policy. What is
needed is a new generation of supercomputers, greater remote sensing
capability,and a larger and more active research community.
In the conventional view, uncertainty is seen as an objectivequantity
whose value can be reduced by investing in more science. While this
may usefully be thought of as one of
several understandings of uncertainty, it is at best simplistic and mis3. The typology of greenhouse "hawks,"
"doves," and "owls"is developed in Michael H.
Glantz, "Politics and the Air Around Us: International Policy Action on Atmospheric Pollution by Trace Gases," in Societal Responses to
Regional Climate Change: Forecasting by Analogy, ed. M. Glantz (Boulder, CO:Westview Press,
1988), pp. 41-42.
SCIENTIFICUNCERTAINTY
leading to think of it as the only or
most important one. Rather than being a cause of controversy,scientific
uncertainty is often a consequenceof
controversy.4This suggests that the
social world is active in the construction and characterization of uncertainty, and if we want to understand
uncertainty, we need to understand
the socialfactorsthat helpto produceit.
UNCERTAINTY,
FALLIBILITY,
ANDINDETERMINISM
The first step in understanding
uncertainty involves distinguishing
it from some related notions with
which it is often confused.5
Uncertainty is often conflated
with fallibility. Fallibility relates to
the fact that we couldbe wrong about
virtually any propositionto which we
give our assent, from the most
homely (forexample, "Iknow how old
I am")to the most exotic (forexample,
"I know how old the universe is").
Fallibility lurks in the backgroundof
scientific claims and moves to the
foreground when new evidence
comes flooding in that suggests that
our previous views about some matter were not just wrong, but deeply
and profoundlywrong. The discovery
of the ozone hole, which was not predicted by any of the atmospheric
models, is one example of this, as is
4. This point is argued forcefully in Brian
L. Campbell, "Uncertainty as Symbolic Action
in Disputes Among Experts," Social Studies of
Science, 15:429-53 (1985).
5. Although I draw the distinctions in a
somewhat different way, my discussion in this
section is indebted to Brian Wynne, "Uncertainty and Environmental Learning: Reconceiving Science and Policy in the Preventive
Paradigm," Global Environmental Change,
2:111-27 (1992).
37
the recognitionof the chronictoxicity
of DDT.6
Fallibility looms large with respect to many health and environmental risks. In some cases, we may
know that various exposures are associated with harms, but we may
have little idea ofwhat causal mechanisms are at work. Although the statistical evidence may be strong
enough for some to attribute causality, even in these cases we may worry
about the fallibility of such claims.
Ourview of the matter may simply be
wrong-not in details, but thoroughly so. We may not even be in a
position to assess the probability of
our being wrong. The fact of our faloften
libility is usually-indeed,
must be-ignored, but it constantly
presents the possibility of bringing
down an entire edifice of knowledge.
Uncertainty arises from ignoring
fallibility.Wetake various features of
a problemas given and focus on other
dimensions. For example, it is widely
agreed that the case for climate
change is weakened by the fact that
we are uncertain about the effects of
clouds on the climate system. The
solution is more intensive study of
cloud formation and effects. But to
identify clouds as an area of uncertainty is to presuppose that our gen-eral knowledge of the climate system
is not uncertain, that the climate
models are basically correct, and so
on. This background knowledge is
"blackboxed"--it is taken as a set of
assumptions from which we proceed
to try to reduce uncertainty.This ap6. For discussion of these cases, see D.
Budansky, "Scientific Uncertainty and the Precautionary Principle," Environment, 33(7):4-5,
43-44 (Sept. 1991).
38
THE ANNALSOFTHE AMERICANACADEMY
proachof taking some propositionsas
fixed while interrogating others is a
fundamental part of scientific practice. Scientific progress would be impossible if every proposition were
problematizedin every investigation.
The general point can be seen from
an everyday example: I discuss selling my bike to a friend. In this context, there is no uncertainty about
whether I own the bike. Weboth take
it as given that this is the case. Of
course, it may be that due to fraud or
forgetfulness I do not own the bike.
But in our discussion, these possibilities are not on the table, and so there
is no uncertainty about whether I
own the bike even though it could
turn out that I do not. Now imagine
a situation in which we are highly
suspicious of each other: it is well
known that I was once convicted of
running a bike theft ring, or that I
suffer from amnesia. When the context is changed in one of these ways,
the problem of uncertainty may
arise. My friend may demand proof
that I really own the bike before she
will continue the discussion with me.
What this homely example shows is
that while we can always be wrong
about (most) things, uncertainty requires particular contexts and social
conditions.
Indeed, this very example has implications for uncertainty about risk.
Uncertainty disappears or is minimized when we have complete trust
in the institution, person, or data set
that is being interrogated. It is magnified or accentuated when there is
mistrust, whether founded on fraud
or other failings.7
Uncertainty should also be distinguished from indeterminacy. Often
what appears to be uncertainty cannot be reduced because there is no
reliable fact of the matter to be
learned that directly bears on improving our beliefs. At least three
sources of indeterminism can be
identified: agency, underdetermination, and categorical relativity.8
Many of the most serious environmental and health problems we face
involve agency.Part of why we do not
know what will happen to global climate in the twenty-first century is
because we do not know how people
will behave in the future. Will they
continueto increase their use of fossil
fuels? Orwill other energy sources be
substituted? Will governments undertake policies to geoengineer climate? Will there be other responses
to early signs of global warming?
These are just a few of the questions
whose answers matter in determining what will happento futureclimate.
Similar questions could be raised
about the effects of tobacco smoke,
the prevalence of HIV,and so on.
The indeterminism that results
fromagency is made worse by the fact
that predictions about human behavior can themselves change the behavior that is being predicted. Consider
a simple case. At 8 a.m. on a warm
summer day, the local radio station
predicts that there will be massive
traffic jams as thousands of people
flock to the beach. The traffic jam
7. For further discussion, see Paul Slovic,
"PerceivedRisk, Trust, and Democracy,"Risk
Analysis, 13(6):675-82(1993).
Disorder ofThings: The Metaphysical Foundations of the Disunity of Science (Cambridge,
8. In addition, some have argued that
indeterminism is a fundamental property of
nature. See, for example, John Dupr6, The
MA:HarvardUniversity Press, 1993).
SCIENTIFICUNCERTAINTY
fails to materialize. Many people
heard the radio broadcast and decided to stay home.
A second source of indeterminism
flows from the underdetermination
of theory by data.9Any particular observation is consistent with an indefinite number of logically distinct theories. For example, the observation
that there are a variety of life forms
is consistent both with evolutionary
theory and creationism. Often we try
to distinguish theories by designing
a crucial experiment, one in which
distinct theories support different
predictions. But there are distinct
theories that cannotbe distinguished
in this way. In such cases, people
often appeal to conceptual concerns
in order to justify the choice of one
theory over another--one theory is
simpler, coheres better with otherbeliefs, and so on. While there may be
grounds for preferringone of two empirically equivalent theories, in such
cases there is no empirical fact of the
matter about which theory is true;
rather, the matter is indeterminate,
for there are no empirical discoveries
that would support one theory at the
expense of the other.
The third source of indeterminism
is even morebasic than the othertwo.
Knowledge claims presuppose categories, but categories are relative.
For example, some people point to
increases in global mean temperatures and extreme climatic events as
evidence of global warming. But why
is global mean temperature a significant category?Why not instead focus
39
on, say, average temperatures? And
why bring together in the single class
of extreme events such diverse phenomena as hailstorms, droughts,
hurricanes, heat waves, cold snaps,
and so on? What are the baselines
from which the claims of increasing
frequency or increasing temperature
are projected?What may appearto be
an increase from a baseline of 50
years ago may appear to be a decrease from a baseline of 500 or 5000
years ago. Of course, stories can be
told about why one form of categorization is better than another; the
point is that empirical investigation
presupposes categories, without being able to justify them empiricallyin
advance.
Rather than being epistemological
problems, fallibility and indeterminism are metaphysical conditions.
Wehave no idea how to overcomeour
fallibility or how to tame those regions of the world that are indeterminate. Uncertainty,on the other hand,
is an epistemological problem. Uncertainty arises fromignoring ourfallibility and winking at indeterminacies. What allows us to do this is a
substratum of conventions, shared
purposes, common contexts, and collective knowledge. Uncertainty is
producednotjust by narrowscientific
mechanisms but also by broad cultural processes. Assertions of uncertainty are not just expressions of our
ignorance but part of what brings
order to our world. Uncertainty implies both the existence of certainty
and the existence of a path from one
to the other. Claims of uncertainty
9. The classic argumentforunderdetermi- reflect and establish epistemological
nation can be foundin W.V.O.Quine,Wordand order and imply a research program
and a way of moving toward closure.
Object(Cambridge:MITPress, 1960).
THE ANNALSOFTHE AMERICANACADEMY
40
THEUSESOFUNCERTAINTY
Sometimes, uncertainty claims are
used directlyin attempts to bring policy debates to closure. For example,
the precautionary principle, which
has been endorsedby various nations
and international bodies, states
(roughly) that if an action or policy
potentially has catastrophic effects,
then we should refrain from undertaking it even if the probabilities are
uncertain.1oOn the other hand, some
argue that unless it is certain that an
action or policy will have harmful
consequences, then it should be permitted. Both views figure in the climate change debate. Greens argue
that since there is a significant
chance that climate change will occur
and have catastrophic consequences,
we should "purchase some insurance" by capping greenhouse gas
emissions. "Browns"argue that unless it is certain that greenhouse gas
emissions will cause catastrophicclimate change, we should not impose
the costs on the economy that capping emissions would entail.
Direct appeals to uncertainty are
rarely effective in bringing policy debates to closure. Instead they often
open the door to the spectacle of duof equal
eling experts-scientists
diahave
who
stature
and
training
what
about
views
metrically opposed
is the case and what ought to be done.
If the experts cannot agree about, for
example, climate change, what is an
ordinary person to think? Rather
than providing a rational means for
resolving epistemological differences, uncertainty reduces science to
10. For further discussion, see Budansky,
"ScientificUncertainty."
just another playground for competing ideologies.
While it is true that scientific uncertainty and the debates that it engenders can be corrosiveto scientific
authority,those who see scientific uncertainty as destructive and delegitimating overlook the fact that virtually all parties to various conflicts
have an interest in maintaining scientific authority.The interest of scientists in maintaining scientific
authority is obvious. But scientists
also benefit from the right amount of
uncertainty. If there is too much uncertainty, an area of research looks
hopeless; if there is too little, research appears not to be needed. The
right amount of uncertainty supports
a call for further research.
Political actors of whatever ideological outlook have an interest in
preserving scientific authority because science can provide a rationalization for decisions that are made on
other grounds. When a policy decision can be presented as dictated by
science, it is a way for a decision
maker to evade responsibility for his
or her choice. A decision backed by
science can be viewed as implied by
the nature of things, not as a decision
for which a leader should be held
accountable. Although political actors have an interest in preserving
scientific authority,they also have an
interest in keeping it in its place. The
optimal role of scientific information
for decision makers is to enable and
structure decisions, not to determine
them.
What I have suggested is that scientific uncertainty mediates between
the closed world of scientific knowl-
SCIENTIFICUNCERTAINTY
edge and the open world of public
policy formation. If what I have said
is correct, the cultural imperative
with respect to scientific uncertainty
is not simply to reduce it but more
generally to manage it. In a recent
article, Shackley and Wynne have
identified some of these management
strategies."1
Quantifyinguncertaintyis one way
of managing it. In 1990, the Intergovernmental Panel on Climate Change
(IPCC)estimated that a carbondioxide doublingwill producean increase
of global mean temperatures of 1.54.0 degrees centigrade.This estimate
summarizes the results of some experiments run on what are regarded
to be the best climate models. The
IPCC estimate does not represent a
probability estimate nor any kind of
normal distribution. Yet specifying
this range as the likely result of a
carbon dioxide doubling sets limits
on the uncertainties, thus making
them more manageable.
Locating uncertainty is another
way of managing it. When a climate
model fails to successfully retrodict a
past climatic condition, this could be
regarded as evidence against the
model. Typically,however,the uncertainties are located not in the model
but in the data that the model manipulates. We are directednot toward
a fundamental rethinking of the
modelbut toward improvingour data
collection.When the uncertaintiesare
located in the data rather than in the
models, they do not threaten the gen-
41
eral project of predicting future climate on the basis of computermodels.
Scheduling reductions in uncertainty is a third way in which uncertainty is managed. The 1990 IPCC
report speaks confidently of reductions in uncertainty that will occuras
a result of better data sets and more
powerful computers. In 1988, the
British Department of the Environment laid out a 25-yearplan for eliminating all of the uncertainties with
respect to future climate.12Of course,
no one knows exactly how these uncertainties will be eliminated. Nonetheless, simply attaching a date to
their elimination appears to make
the problems more tractable.
IMPROVINGDECISIONMAKING
Despite the fact that scientific uncertainty plays a functional role in
our public decision-making processes, many people are unhappy about
how we make decisions that have important scientific dimensions. The
core of the unhappiness, I believe, is
that the gap between science and policy seems unnecessarily wide. As a
society,we have a large investment in
science, yet science seems to influence policy only indirectly. Science
and policy can be brought into closer
contact,but there is a price that must
be paid. Here are some positive suggestions for how science can be
brought into closer contact with policy questions.
First, greater attention can be
paid to problem definition at the be11. Simon Shackley and Brian Wynne, ginning of a decision-makingprocess.
"RepresentingUncertainty in Global Climate When policy problems are not
clearly
ChangeScienceandPolicy:Boundary-Ordering defined and
it
is
characterized,
quite
Devices and Authority," Science, Technology
and Human Values (in press).
12. Ibid., p. 25 (in manuscript).
42
THE ANNALSOFTHE AMERICANACADEMY
unclear what scientific informationis
relevant to bringing them to closure.
Better problem definition involves
being clear not only about what questions are being asked but also about
the context in which they are asked
and the purposesthat answersto these
questions are supposed to serve. The
debate over climate change policy is
an example of how things can go
wrong when there is little agreement
about what question is being asked.
Some people claim that it is uncertain whether emitting greenhouse
gases will change climate; others
seem to deny this. In some cases, they
are not really disagreeing. Both parties to the dispute may agree that, for
the purposes of counting as scientific
knowledge, the proposition is uncertain. Moreresearch needs to be done,
data collected, and so forth. But those
who seem to deny that there is significant uncertainty are often claiming not that there is no scientific uncertainty but that there is no
uncertainty for the purposes of public
decision making. In their view, the
risk of climate change is known to be
great enough, and the costs of mitigation and prevention are low enough,
that some "no regrets" strategies
ought to be pursued. This is an example of a case in which it is clear that
the scientific data may rightfully be
regarded as uncertain for some purposes but not for others."3
Second,variousreformsin the practice of science would help in bringing
scientific informationinto closer con13. These suggestions are more fully developed in Charles Herrick and Dale Jamieson,
"The Social Construction of Acid Rain," Global
Environmental Change, 5(2):105-12 (May
1995).
tact with public decision making.14
Science, as it is practicedin American
society, is an elite institution, to a
great extent self-governing,with primary allegiance to its own internal
values. While many people have access to the deliverances of science,
very few people are involved in the
production of science, and scientists
themselves are overwhelmingly
white, male, and upper middle class.
Finally, there are various reforms
in our public decision-making processes that would also help to bring
science into greater contact with policy.As things now stand, science and
science policy are scattered throughout the federal government. We are
one of the few industrialized nations
that does not have a cabinet-level
department of science and technology.Moreover,the adversarial way in
which policy debate is conducted in
this country may also have the effect
of marginalizingor needlessly problematizing scientific information. Scientific institutions are in many ways
authoritarian and directed toward
the creation of consensus and thus
are often at odds with the prevailing
values of policy debate.
Broad changes in the areas I have
identified would bring scientific information into closer contact with
policy,but as a result, science would
become less autonomous, and public
decision making might become more
technocratic.Even if it were thought
that this pricewas not too high to pay,
the effect that science would have on
14. I have discussed some of these reforms
in "What Society Will Expect from the Future
Research Community," Science and Engineering Ethics, 1(1):73-80 (1995).
SCIENTIFICUNCERTAINTY
policy decisions would still remain
limited.
One reason the role of science
would remain limited is that our
most important public policy decisions involve questions of value that
cannot be addressedby science.Asecond reason why science would continue to have a limited role relates to
the prevailing cultural attitudes that
frame our decision-makingpractices.
We are living in a time in which citizens are deeply insecure about their
own futures and those of their children and have very little trust in institutions of any sort. In such an atmosphere, the bonds of community
are strained and the willingness to
make trade-offs is limited. Against
such a background,science, however
43
certain, is of limited effectiveness in
shaping people'sview of the world.
CONCLUDINGREMARKS
In this article, I have sketched a
notion of scientific uncertainty that is
in many ways different from the prevailing view. Scientific uncertainty is
not simply an objectivevalue that can
be reduced by science alone. Rather,
scientific uncertainty is constructed
by both science and society in order
to serve certain purposes. Recognizing the social role of scientific uncertainty will help us to see how many
of our problemsabout risk are deeply
cultural and cannotbe overcomesimply by the application of more and
better science.